Status page best practices: what to show, what to write and what to avoid
How to decide which services a status page lists, how to word incidents and maintenance, where to host it, and the mistakes that make customers stop trusting it.
· 6 min read · By the Spot Downtime team
Setting up a status page takes ten minutes. Making one that people believe takes a few decisions you only notice after your first bad day. This post is about those decisions: what to list, how to word things, and the habits that keep the page trustworthy. For the step-by-step setup, see how to set up a status page.
1. List what customers use, not what you run
Your infrastructure has a database, a queue, three workers and a cache. Your customers have "I can't log in" and "my invoice didn't send". Name components after what the customer does:
- Good: Website, Sign-in, API, Checkout, Email notifications, Mobile app.
- Poor:
pg-primary-eu,redis-2,worker-pool-b.
Aim for roughly four to ten components. Fewer and a partial outage has nowhere to show up. More and nobody can find theirs. If your product runs in separate regions that customers choose between, list the region as part of the name so a visitor can tell whether it affects them.
2. Let monitoring drive the status
A page updated by hand is only as fast as the person who remembers to update it, and that person is busy fixing the outage. Connect each component to a monitor so it turns red when the check fails. Then a human adds the explanation, instead of first having to notice.
In Spot Downtime the page updates itself from the monitors you tick, shows 90 days of daily uptime bars and refreshes every minute; the details are on the status pages feature page. Choose monitors that reflect the real experience: a full-flow API check for sign-in is a better signal for "Sign-in" than a ping to the server.
Automatic is not the same as silent
3. Host it away from your own stack
If the status page lives on the same servers, domain or DNS as your product, the day your product goes down is the day the status page goes down with it. Use a separate provider, and if you use a custom domain, remember that its DNS is a dependency too. A hosted page under a different domain or subdomain, with its own certificate, is the safest default. A custom domain is worth it for trust; just keep the record monitored. See the custom domains guide.
4. Write updates like a person
An incident update has four jobs: say what is affected, say what you are doing, say when you will update next, and say when it is over. Keep the sentences plain.
- Investigating. "Some customers can't sign in. We are looking into it. Next update by 14:30 UTC."
- Identified. "A bad configuration change is stopping sign-ins. We are rolling it back."
- Monitoring. "The rollback is done and sign-ins work again. We are watching to be sure."
- Resolved. "Sign-in has been working normally since 14:52 UTC. A full write-up will follow."
Promise a next-update time and keep it, even when the update is "no change". Silence between updates is what customers read as "they don't know either". More examples and wording are in our guide to writing incident status updates.
5. Say what you do not know, and do not guess
- Skip a root cause until you have one. "We are investigating" is honest; a wrong cause that you retract later is not.
- Do not promise a fix time you cannot control. Give the next update time instead.
- Do not blame a vendor in the first hour. If a provider is at fault, say which service is affected and leave the rest for the review.
- Leave out hostnames, internal tool names, IP addresses and error text. Customers do not need them, and attackers do.
6. Be accurate about severity
Use the levels for what they mean. Degraded or partial outage means something works for some people or is slow; major outage means customers cannot use it. Marking everything "major" trains people to ignore the colour, and marking a real outage "degraded" makes the page look like spin. If a component is not working for the customers who depend on it, it is down, however healthy the servers look.
7. Announce maintenance in advance
- Post it at least a few days ahead for anything customers will notice, with the start time, expected length and what is affected.
- State times in UTC, or with an explicit time zone, not just your local time.
- Mark the window so planned downtime is not counted as an outage in your numbers, and say so.
- Close it out when it is done, and extend it visibly if it overruns.
On the technical side, serve a 503 with a Retry-After header while you are down; see 503 maintenance mode and Retry-After.
8. Do not edit the past
The history is the proof. Do not delete an incident or shorten an outage to improve a percentage. Customers compare what they experienced with what you posted, and one mismatch undoes months of green bars. If you got an update wrong, add a correction rather than rewriting the old one. For what the numbers mean, see what 99.9% uptime really means.
9. Give people a way to hear without checking
- Offer email subscriptions, so customers hear about an incident without refreshing the page.
- Link the page from your app's footer, error pages, support signature and help centre. People who can find it use it.
- Optionally show a live status pill on your own site with the status widget.
10. Close the loop with a review
For anything customers noticed, publish a short review once you understand it: what happened, how long, what you changed. It takes an hour and does more for trust than any design. A template is in our post-incident review template, and the first message you send to customers is covered in what to tell customers when your site is down.
A short checklist
- Components are named for what customers do, four to ten of them.
- Each is driven by a monitor, with a person adding the words.
- The page does not share servers, DNS or a certificate with your product.
- Updates name the impact, the action and the next update time.
- Severity levels mean what they say, and history is never edited.
- Maintenance is announced ahead, in UTC.
- Customers can subscribe, and can find the page in one click.
Get new guides by email
Practical posts on monitoring, outages, error fixes and on-call, when they're published. No spam.
Keep reading
- Status pages · IncidentsHow to set up a status page customers actually trustWhat a good status page shows, why it should update itself from monitoring, and how to add your own domain, email subscriptions and maintenance notices.October 2, 2026 · 4 min read
- Incidents · Status pagesWebsite down message to customers: templatesCopy-ready messages for when your website or app is down: a status page update, banner, email, social post, support auto-reply, maintenance page and the all-clear.November 2, 2026 · 6 min read